We combine network theory and the knowledge-based approach in organization theory to predict to what extent dynamics in various loci of field networks influence innovation outcomes of research and development (R&D) consortia differently. Using unique quantitative and qualitative data on publicly funded R&D consortia involving a mix of public and private organizations active in the Dutch water sector from 1982 to 2004, we demonstrate the relevance of different loci of dynamics in field networks for innovation outcomes. Both clusters in field networks and connectors between these clusters can range from stable to dynamic, resulting in four configurations of field network dynamics. The ‘dynamic connector-stable cluster’ and ‘stable connector-dynamic cluster’ configurations facilitate superior innovation outcomes, relative to the other configurations, by blending the benefits of dynamics versus stability (i.e. knowledge diversity versus mutual understanding) with the advantages of clusters versus connectors (i.e. organizing capacity versus knowledge access). Our findings contribute to our understanding of how dynamics in specific field network loci impact innovation outcomes of R&D consortia. This study also reveals the endogenous nature of network stability, which carries important implications for policymakers that wish to influence the effectiveness of public-private R&D collaboration.